Four consecutive coral bleaching events in the Northern Persian Gulf: 2014–2017

In the past few decades, global coral reefs have been threatened by a wide range of human activities such as accelerated industrialization, urbanization, agriculture, and natural phenomena including storms and biological stressors [1-4]. However, global warming-induced coral bleaching and mortalities due to fast increases in temperatures that pass coral species’ thermal tolerance have become the most prominent concern [5-7]. Such temperature anomalies have destroyed coral reefs worldwide, particularly during the four-year period from 2014 to 2017 [6-12]. For example, mass bleaching happened in the Great Barrier Reef, Australia, in 2016 and 2017 [7,8] and in the central Indian Ocean in 2015 and 2016 [13] including the Persian Gulf [14-18].

In the northern Persian Gulf, where bleaching threshold temperatures are at least 1.5℃ to 2.5℃ lower than the southern part, bleaching and mortality events were reported in 2012 [29], 2007, and 2017, with severe long-term consequence [14,16,18].
Here, we report four back-to-back bleaching and mortality events at four sites on two Iranian islands, Hormuz and Larak Islands, during the period 2014-2017. This is to our knowledge the fi rst time in the history of modern coral reefs that some reefs have experienced four consecutive bleaching events.

Study sites
Two major reef sites of Hormuz Island, located in the south (known as the Red Soil, H-RS; 27°01'N, 56°27'E) and in the east (H-E; 27°03'N, 56°30'E) and two sites on Larak Island, at the north of the island (L-N; 26°88'N, 56°35'E) and at the southwest (L-SW; 26°49'N, 56°18'E) were chosen. The 2012 mass bleaching was previously studied and recorded at both the Hormuz Island and L-SW site of Larak Island [29].
The composition of coral taxa was different among sites and faced various stressors in the past [28,29,32,33]. Turbidity is a natural characteristic of all sites, particularly at the sites of Hormuz Island. The most turbid site is H-RS, where horizontal visibility declines to < 2 meters. In addition, water currents at the H-E site are very strong so that it even made it diffi cult for us to conduct this research. The depth of corals at all sites was less than 10 m.

Survey method
At each site in each year, eight belt transects of 20m*1 m were randomly selected and photographed/recorded. The coral colonies inside each transect were identifi ed to the genus level.
Each colony was categorized based on its bleaching status as unbleached or Healthy (H) if there was no sign of bleaching, moderately bleached (M) if it was bleached < 50%, or severely bleached (S) if it was bleached for > 50%. Each coral colony was identifi ed based on its mortality status as one of the following categories: 1) 0%-20%; 2) 21%-40%; 3) 41%-60%; 4) 61%-80%, and 5) 81%-100%. Mean abundance of coral colonies at each site was calculated based on the average number of coral colonies per transect (n= 8).

Statistical analysis
All statistical analysis were performed in RStudio software (RStudio, Boston, United States). Bleaching rate, mortality rate, and abundance of corals at the study sites in 2014-2017 were assessed using a three-way ANOVA. For all the aforementioned parameters, data were averaged per site (n = 8 transects).
The normality of the residuals was verifi ed with a Shapiro-Wilk's test and the homogeneity of variances was tested using Levene's test. A Tukey-adjusted pairwise comparison between years or between bleaching/mortality categories at each year at each site was applied as a post hoc test when the ANOVA analysis showed a signifi cant effect.

Results
There were signifi cant differences among the three bleaching categories in each year and for each bleaching category among years for each site (Table 1). There were four distinguishable bleaching patterns observed at the sites ( Figure 1, Table 2). At site H-E, about 69% of the coral colonies were unbleached in 2014 while in 2017 the percent of unbleached coral colonies dropped to < 20%. In 2015, the percent of moderately bleached corals rose from < 20% in 2014 to about 78% and remained almost constant in 2016 and 2017. However, the percent of severely bleached colonies, which was < 15% in the fi rst there years, signifi cantly increased in 2017. At site H-RS, in all four years, the percent of severely bleached colonies was signifi cantly higher than the other two categories and the percent of moderately bleached coral colonies was signifi cantly higher than that of healthy (unbleached) corals in 2015 and 2017, and numerically higher in 2014 and 2016. None of the bleaching categories showed a difference between years. At site L-N, the percent of unbleached coral colonies was about 5% in 2014 and reached 0 in the following years. In 2014, there was no signifi cant difference between the percent of moderately and severely bleached corals (Table 1). However, in the following two years, moderately bleached corals reached a signifi cantly higher proportion and in 2017 the percent of severely bleached corals increased to about 78%, which was signifi cantly higher than the percent of moderately bleached corals. The bleaching pattern at L-SW was similar to that of L-N with one major difference: the differences between the proportion of moderately and severely bleached corals in 2014, 2015, and 2016 were non-signifi cant.
Bleaching intensity for the majority of coral genera under consecutive bleaching events signifi cantly increased in 2017 compared to previous years so that coral genera that were mainly moderately bleached in the fi rst three years were severely bleached in 2017 ( Figure S1). Dipsastraea at the site H-RS, whose whole population was severely bleached in the previous three years, showed a signifi cant reduction in the percent of severely bleached corals without any increase in the percent unbleached corals ( Figure S1).
There were signifi cant differences among the mortality categories in each year and for each bleaching category among years for each site (Table 3). Mortality remained low at H-E  Table 2.  and H-RS with no signifi cant difference between mortality categories in each year or for each mortality category between years ( Figure 2, Table 4). Similarly, mortality rates at L-N and L-SW were low with no signifi cant difference except in 2017, when abundance of colonies with 81-100% mortality was signifi cantly higher than the other categories ( Table 4).
The coral genera mainly showed limited mortality and many of them showed signs of mortality in just one year ( Figure S2). For example, almost all Acropora colonies at the site L-N that were recorded with minimal mortality in the fi rst three years showed 81%-100% mortality in 2017. Acropora at the site L-SW and Porites at the site L-N showed signifi cant increases in percent coral colonies that were severely bleached from 2014 toward 2017 ( Figure S2).  Figure S3). Porites at the site L-N declined 66% in 2017 compared to 2014. Acropora at the site L-N and Porites at the site L-SW showed signifi cant increases in 2015 and 2016 compared to the previous years, respectively; however, they both declined by approximately 50% in the following year (signifi cantly for Acropora and non-signifi cantly for Porites; Figure S3). The abundance of Dipsastraea at the site L-SW showed signifi cant increases in 2015 and 2016 compared to previous years, but a non-signifi cant decline in 2017 ( Figure  S3).

Discussion
To the best of our knowledge, this is the fi rst time in the history of modern coral reefs that four consecutive bleaching events have been reported. It must be taken as a serious warning message for all the world's coral reefs because fi rst, the Persian Gulf has been considered a coral reef refugium, where corals could survive climate change by 2100 [21,22]. Secondly, these corals are known to encompass some of the most thermotolerant reef-building corals [19,34] and associated endosymbiotic algae [35][36][37][38] and face the highest SST records in the world [14,39]. Third, seawater turbidity, which may reduce the severity of bleaching [40][41][42][43][44][45] is an intrinsic characteristic of the Persian Gulf including our study sites, particularly on Hormuz Island [28,29]. Fourth, strong water currents that were suggested to ameliorate the negative effects of thermal stress on corals [45][46][47], were present at the site H-E. Conversely, our data question such predictions and show that neither the thermotolerance of the corals and their symbionts nor the natural turbidity and strong water currents could protect coral reefs of the Persian Gulf. It was suggested  that the Persian Gulf will be the last place where coral reefs will face annual severe bleaching [21]. However, our study warns that the coral reefs of the Persian Gulf may be among the fi rst reefs in the world to disappear. Mass coral bleaching events have been frequently reported from the Persian Gulf in the past three decades [15,24,27,30,31]. Three back-to-back bleaching events have happened before this in the southern Persian Gulf in 2010, 2011, and 2012 [27]. However, it seems that the northern Persian Gulf is becoming a major bleaching hotspot.
The four consecutive bleaching events in the northern Persian Gulf happened just one year after the 2012 massive bleaching that resulted in the bleaching of 84% of the corals [29]. For example, 100% of the corals at H-E were to some This confi rms previous studies that suggested consecutive bleaching events can lead to coral reef degradation because they do not give the bleached corals enough time to recover and produce larvae in order to return the reefs to their original, pre-bleached state [27,48]. Bleaching events may have longterm dramatic impacts on coral reef ecosystems and services that can be observed years later [14,[49][50][51]. (against 73%-86% at other sites). One reason could be the fact that the dominant coral genus at H-E is the massive Porites, which accounted for > 85% of the coral cover [28]. In other sites, massive Porites was not the dominant genus. Massive Porites species are among the most tolerant reef-building coral species against elevated temperatures [52,53]. Some studies suggested that bleaching events may increase the thermotolerance capacity of the corals via acclimation/adaptation [54][55][56].
Even short-term winners may become losers over time [52].
Therefore, although some tolerant colonies may survive even four back-to-back bleaching events, by increasing the number of severe bleaching events all coral taxa may be losers.

Conclusion
In conclusion, our study shows that even the most thermotolerant coral reefs of the world can face back-to-back bleaching events that lead to increased coral mortality and reduction in coral abundance at some sites. The signifi cant increases in the frequency of Acropora coral colonies with 81%-100% mortality at the sites of Larak Island in 2017 despite being constantly minimal in the fi rst three years highlights the fact that we cannot rely on the physiological acclimation/ adaptation of corals or natural phenomena such seawater turbidity to save coral reefs. The repeated bleaching events did not seem to lead to increased tolerance of the corals of the study sites, except for Dipsastraea at the southwest of Larak Island that managed to signifi cantly increase its population in 2015 and 2016. However, the non-signifi cant 27% reduction in its abundance in 2017, when almost all colonies were severely bleached suggest that these results must be interpreted with caution. In particular that, for Dipsastraea at the north of Larak Island that managed to survive in the fi rst three years, > 66% of its colonies showed 81%-100% mortality in 2017.
Unfortunately, there is no effective management strategy to protect global coral reefs from climate change-induced thermal stress [7]. Therefore, as previously suggested, it seems the only way to prevent reef-building corals from reaching extinction is to substantially reduce greenhouse gas emissions, including carbon dioxide.

Author contributions
JK designed the experiment, collected the data, wrote the paper, PTK collected the data, SHK collected the data, and FG analyzed the data.